Published on Web 03/06/2009
Plasmonic Cu2-xS Nanocrystals: Optical and Structural
Properties of Copper-Deficient Copper(I) Sulfides
Yixin Zhao,† Hongcheng Pan,†,‡ Yongbing Lou,† Xiaofeng Qiu,† JunJie Zhu,*,‡ and
Clemens Burda*,†
Center for Chemical Dynamics and Nanomaterials Research, Department of Chemistry, Case
Western ReserVe UniVersity, 10900 Euclid AVenue, CleVeland, Ohio 44106, and Key Laboratory
of Analytical Chemistry for Life Science Department of Chemistry, Nanjing UniVersity,
Nanjing 210093, P. R. China
Received July 20, 2008; E-mail: [email protected]; [email protected]
Abstract: Cu2-xS (x ) 1, 0.2, 0.03) nanocrystals were synthesized with three different chemical methods:
sonoelectrochemical, hydrothermal, and solventless thermolysis methods in order to compare their common
optical and structural properties. The compositions of the Cu2-xS nanocrystals were varied from CuS
(covellite) to Cu1.97S (djurleite) through adjusting the reduction potential in the sonoelectrochemical method,
adjusting the pH value in the hydrothermal method and by choosing different precursor pretreatments in
the solventless thermolysis approach, respectively. The crystallinity and morphology of the products were
characterized by X-ray diffraction (XRD) and transmission electron microscopy (TEM), which shows that
most of them might be of pure stoichiometries but some of them are mixtures. The obtained XRDs were
studied in comparison to the XRD patterns of previously reported Cu2-xS. We found consistently that under
ambient conditions the copper deficient Cu1.97S (djurleite) is more stable than Cu2S (chalcocite). Corroborated
by recent computational studies by Lambrecht et al. and experimental work by Alivisatos et al. This may
be the reason behind the traditionally known instability of the bulk Cu2S/CdS interface. Both Cu2S and the
copper-deficient Cu1.97S have very similar but distinguishable electronic and crystal structure. The optical
properties of these Cu2-xS NCs were characterized by UV-vis spectroscopy and NIR. All presented Cu2-xS
NCs show a blue shift in the band gap absorption compared to bulk Cu2-xS. Moreover the spectra of these
Cu2-xS NCs indicate direct band gap character based on their oscillator strengths, different from previously
reported experimental results. The NIR spectra of these Cu2-xS NCs show a carrier concentration dependent
plasmonic absorption.
Introduction
Since the discovery of CdS/Cu2S heterojunction solar cells
in 1954,1 copper sulfides have drawn a lot of attention as an
important component in photovoltaic cells.2,3 Copper sulfides
(Cu2-xS) can vary their band gap and structure with varying
stoichiometries. The stoichiometric factor (2-x) in Cu2-xS varies
in a wide range between 1 and 2, from the Cu-rich Cu2S
(chalcocite) to CuS (covellite). Intermediate phases include
Cu1.97S (djurleite), Cu1.8S (digenite), Cu1.4S (anilite) and others
have been produced. These Cu2-xS nanocrystals can be used as
p-type semiconductor due to the copper vacancies in the lattice,
which is the reason for their use in optoelectronic devices.4,5
Most recent theoretical work has studied the electronic structure
†
Case Western Reserve University.
Nanjing University.
(1) Reynolds, D. C.; Leies, G.; Anter, L. L.; Margurber, R. E. Phys. ReV.
1954, 96, 533–534.
(2) Chopra, K. L.; Das, S. R. Thin Film Solar Cells; Plenum Press: New
York, 1983.
(3) Neville, R. C. Solar Energy ConVersion: The Solar Cell; Elsevier:
Amsterdam, 1995.
(4) Partain, L. D.; Mcleod, P. S.; Duisman, J. A.; Peterson, T. M.; Sawyer,
D. E.; Dean, C. S. J. Appl. Phys. 1983, 54, 6708–6720.
(5) Lou, Y. B.; Samia, A. C. S.; Cowen, J.; Banger, K.; Chen, X. B.;
Lee, H.; Burda, C. Phys. Chem. Chem. Phys. 2003, 5, 1091–1095.
‡
10.1021/ja805655b CCC: $40.75  2009 American Chemical Society
of these interesting different copper sulfides.6 The results showed
that there is no evidence for an indirect band gap in hexagonal
Cu2-xS, which is distinctly different from previous reports.4 Due
to the attractive applications and an attempt to better understand
these Cu2-xS, a lot of methods have been developed for the
synthesis of nanostructured copper sulfides.5,7-15 Lou et al. used
a single source precursor method to decompose Cu(S2CNEt2)2
in a TOP/TOPO/TOPS solvent system to obtain spherical Cu1.8S
(digenite) nanocrystals.5,16-18 Klimov et al. used a topochemical
method to obtain copper sulfide nanoparticles.19 Ghezelbash et
(6) Lukashev, P.; Lambrecht, W. R. L.; Kotani, T.; van Schilfgaarde, M.
Phys. ReV. B 2007, 76, 195202/1–195202/14.
(7) An, C. H.; Wang, S. T.; He, J.; Wang, Z. X. J. Cryst. Growth 2008,
310, 266–269.
(8) Chen, L.; Chen, Y. B.; Wu, L. M. J. Am. Chem. Soc. 2004, 126, 16334–
16335.
(9) Ding, Y.; Liu, X.; Guo, R. Mater. Res. Bull. 2008, 43, 748–758.
(10) Du, X. S.; Yu, Z. Z.; Dasari, A.; Ma, J.; Meng, Y. Z.; Mai, Y. W.
Chem. Mater. 2006, 18, 5156–5158.
(11) Kuzuya, T.; Itoh, K.; Sumiyama, K. J. Colloid Interface Sci. 2008,
319, 565–571.
(12) Kuzuya, T.; Tai, Y.; Yamamuro, S.; Sumiyama, K. Sci. Technol. AdV.
Mater. 2005, 6, 84–90.
(13) Lim, W. P.; Wong, C. T.; Ang, S. L.; Low, H. Y.; Chin, W. S. Chem.
Mater. 2006, 18, 6170–6177.
(14) Zhang, H. T.; Wu, G.; Chen, X. H. Langmuir 2005, 21, 4281–4282.
(15) Zhang, J. Z. J. Phys. Chem. B 2000, 104, 7239–7253.
J. AM. CHEM. SOC. 2009, 131, 4253–4261
9
4253
ARTICLES
Zhao et al.
al. synthesized copper sulfide nanocrystals by a high temperature
solution phase method.20 Other methods like the hydrothermal
approach and the aqueous colloidal capping method have also
been introduced into the synthesis of copper sulfide nanoparticles.8 Considerable efforts have been devoted to the shape
controlled synthesis of nanostructured copper sulfides, various
morphologies of copper sulfides have been fabricated such as
nanorods,21-24 nanotubes4,25 nanofibers,26 nanowhiskers,27
nanoflakes,28 nanowalls,29 nanocages,30 flowerlike,31,32 hollow
spheres,33,34 and quadrates with varying degree of stoichiometric
control.35 The band gap of these Cu2-xS exhibits stoichiometry
dependence. An increase in the band gap occurs with an increase
of the x value in bulk copper sulfides (Eg ) 1.2 eV for Cu2S,
1.5 eV for Cu1.8S and 2.0 eV for CuS).36-38 Compositional
control of copper sulfide nanocrystals is a way to tune the
optoelectrical properties of Cu2-xS-based materials. Reijnen and
co-workers reported the fabrication of Cu2-xS thin films by using
atomic layer deposition.37,39 Xu et al. presented a templateassisted synthesis of well-defined Cu2-xS mesocages by using
Cu2O crystals as sacrificial templates. The compositions of the
Cu2-xS mesocages could be adjusted from Cu2S to Cu1.75S by
control of the reaction conditions from a nitrogen to an air
atmosphere.40
The difficulties in composition control, product impurities,
and critical reaction conditions are the common challenges of
these mentioned methods. Here, we report the novel convenient
(16) Yin, M.; Wu, C. K.; Lou, Y. B.; Burda, C.; Koberstein, J. T.; Zhu,
Y. M.; O’Brien, S. J. Am. Chem. Soc. 2005, 127, 9506–9511.
(17) Lou, Y. B.; Yin, M.; O’Brien, S.; Burda, C. J. Electrochem. Soc. 2005,
152, G427–G431.
(18) Lou, Y. B.; Chen, X. B.; Samia, A. C.; Burda, C. J. Phys. Chem. B
2003, 107, 12431–12437.
(19) Klimov, V.; Bolivar, P. H.; Kurz, H.; Karavanskii, V.; Krasovskii,
V.; Korkishko, Y. Appl. Phys. Lett. 1995, 67, 653–655.
(20) Ghezelbash, A.; Korgel, B. A. Langmuir 2005, 21, 9451–9456.
(21) Kalyanikutty, K. P.; Nikhila, M.; Maitra, U.; Rao, C. N. R. Chem.
Phys. Lett. 2006, 432, 190–194.
(22) Roy, P.; Srivastava, S. K. Mater. Lett. 2007, 61, 1693–1697.
(23) Larsen, T. H.; Sigman, M.; Ghezelbash, A.; Doty, R. C.; Korgel, B. A.
J. Am. Chem. Soc. 2003, 125, 5638–5639.
(24) Sigman, M. B.; Ghezelbash, A.; Hanrath, T.; Saunders, A. E.; Lee,
F.; Korgel, B. A. J. Am. Chem. Soc. 2003, 125, 16050–16057.
(25) Wu, C.; Yu, S. H.; Chen, S.; Liu, G.; Liu, B. J. Mater. Chem. 2006,
16, 3326–3331.
(26) Du, W.; Qian, X.; Ma, X.; Gong, Q.; Cao, H.; Yin, J. Chemistry 2007,
13, 3241–3247.
(27) Puspitasari, I.; Gujar, T. P.; Jung, K. D.; Joo, O. S. Mater. Sci. Eng.,
B 2007, 140, 199–202.
(28) Zhang, H. T.; Wu, G.; Chen, X. H. Mater. Chem. Phys. 2006, 98,
298–303.
(29) Feng, X.; Li, Y.; Liu, H.; Cui, S.; Wang, N.; Jiang, L.; Liu, X.; Yuan,
M. Nanotechnology 2007, 18, 145706.
(30) Xu, H. L.; Wang, W. Z.; Zhu, W.; Zhou, L. Nanotechnology 2006,
17, 3649–3654.
(31) Ji, H.; Cao, J.; Feng, J.; Chang, X.; Ma, X.; Liu, J.; Zheng, M. Mater.
Lett. 2005, 59, 3169–3172.
(32) Wu, Z. C.; Pan, C.; Yao, Z. Y.; Zhao, Q. R.; Xie, Y. Cryst. Growth
Des. 2006, 6, 1717–1719.
(33) Zhu, H.; Ji, X.; Yang, D.; Ji, Y.; Zhang, H. Microporous Mesoporous
Mater. 2005, 80, 153–156.
(34) Ni, Y.; Liu, H.; Wang, F.; Yin, G.; Hong, J.; Ma, X.; Xu, Z. Appl.
Phys. A: Mater. Sci. Process. 2004, 79, 2007–2011.
(35) Liu, Y.; Qin, D.; Wang, L.; Cao, Y. Mater. Chem. Phys. 2007, 102,
201–206.
(36) Nair, M. T. S.; Guerrero, L.; Nair, P. K. Semicond. Sci. Technol. 1998,
13, 1164–1169.
(37) Reijnen, L.; Meester, B.; Goossens, A.; Schoonman, J. Chem. Vap.
Deposition 2003, 9, 15–20.
(38) Nascu, C.; Pop, I.; Ionescu, V.; Indrea, E.; Bratu, I. Mater. Lett. 1997,
32, 73–77.
(39) Reijnen, L.; Meester, B.; deLange, F.; Schoonman, J.; Goossens, A.
Chem. Mater. 2005, 17, 2724–2728.
(40) Jiao, S.; Xu, L.; Jiang, K.; Xu, D. AdV. Mater. 2006, 18, 1174–1177.
4254
J. AM. CHEM. SOC.
9
VOL. 131, NO. 12, 2009
chemical synthesis of Cu2-xS (x ) 1, 0.2, 0.03) nanocrystals
with three different convenient and environmental benign
techniques based on sonoelectrochemical,41,42 hydrothermal,43
and solventless thermolysis methods.23,44 The presented methods
can produce Cu2-xS (x ) 1, 0.2, 0.03) nanocrystals and were
used to elucidate some of the electronic and structural properties
of the currently presented and some recently reported copper
sulfide nanoparticles. This paper aims to address some of the
existing uncertainties about the physical-chemical properties of
Cu2-xS nanocrystals. Namely, the challenges in assignment of
their X-ray and optical spectra are identified and clarified in
the present work.
Experimental Section
Sonoelectrochemical Method. The room-temperature sonoelectrochemical synthesis of copper sulfide nanocrystals (NCs) is based
on constant-potential electrodeposition accompanied with simultaneous ultrasonication in aqueous solution.41,42 Cupric sulfate (33
mM) was first dissolved in water and then mixed with a sodium
thiosulfate solution (16.5 mM), and the citric acid concentration
was 87 mM. VCX-500 ultrasonic processors from Sonics and
Materials Inc. (Ti horn, 0.5 in. in diameter, 20 kHz) worked as
ultrasound source and a CHI 600B electrochemistry workstation
was used to control and supply a constant potential. The solution
was sonicated in the presence of a constant potential in the range
0 to -1.5 V (relative to Ag/AgCl satd KCl electrode) for ∼1 h,
the CuS, Cu1.8S, and Cu1.97S NCs were produced at potentials 0,
-0.6, and -1.2 V, respectively. After the reaction, the resulting
nanocrystal suspension was centrifuged and the precipitate was
washed and centrifuged three times with distilled water and ethanol.
Hydrothermal Method. In a three-neck flask, 1 mmol of
CuSO4 · 5H2O was dissolved in 95 mL of distilled water and stirred
for 15 min. Mercapto-propionic acid (MPA, 135 µL) was added
into the solution under stirring. The yellow precipitate was obtained
after several minutes, indicating the formation of a Cu-MPA
complex. The pH value of solution was adjusted from 2 to 10 by
adding 1 M NaOH solution. Then the solution was heated to 100
°C and 5 mL of 0.2 M Na2S2O3 aqueous solution was added. The
solution was refluxed at 100 °C for 7 h to produce a dark-brown
precipitate. CuS and Cu1.97S NCs were produced at pH 2 and 5,
respectively. After the reaction, the resulting nanocrystal suspension
was centrifuged and the precipitate was washed and centrifuged
three times with distilled water and ethanol.
Solventless Thermolysis. In a three-neck flask, 1 mmol of
CuSO4 · 5H2O was dissolved in 100 mL of distilled water and stirred
for 15 min. MPA (135 µL) was added into the solution under
stirring. The yellow precipitate was separated by centrifuging and
washed with water and acetone. This washing procedure was
repeated three times with water and three times with acetone. The
resulting precipitate was post-treated by the following two procedures.
(A) The wet precipitate was dried in air at room temperature for
3 days. The resulting green yellow powder was heated in nitrogen
atmosphere at 200 °C for 60 min and cooled to room temperature.
A dark-brown Cu1.8S powder was obtained.
(B) The wet precipitate was nitrogen-flooded for 30 min to obtain
a dried waxlike gel. The resulting dried gel was heated in nitrogen
atmosphere to 200 °C for 60 min and cooled to room temperature.
A brown Cu1.97S powder was obtained.
The resulting powder was washed and centrifuged three times
with water and ethanol.
(41) Qiu, X. F.; Burda, C.; Fu, R. L.; Pu, L.; Chen, H. Y.; Zhu, J. J. J. Am.
Chem. Soc. 2004, 126, 16276–16277.
(42) Qiu, X. F.; Lou, Y. B.; Samia, A. C. S.; Devadoss, A.; Burgess, J. D.;
Dayal, S.; Burda, C. Angew. Chem., Int. Ed 2005, 44, 5855–5857.
(43) Roy, P.; Srivastava, S. K. Cryst. Growth Des. 2006, 6, 1921–1926.
(44) Sigman, M. B.; Ghezelbash, A.; Hanrath, T.; Saunders, A. E.; Lee,
F.; Korgel, B. A. J. Am. Chem. Soc. 2003, 125, 16050–16057.
Plasmonic Cu2-xS Nanocrystals
Figure 1. Powder X-ray diffraction patterns of CuS (A), Cu1.8S (B), and
Cu1.97S (C) nanocrystals prepared at potentials of 0, -0.6, and -1.2 V,
respectively. The bars are from JCPDS standards for hexagonal CuS[792321], hexagonal Cu1.8S[47-1748], and orthorhombic Cu1.97S[20-0365].
ARTICLES
Figure 2. Powder X-ray diffraction patterns of copper sulfide nanocrystals
prepared by the hydrothermal method at pH 2 (A), 3 (B), and 5 (C). The
bars are from standard JCPDS files of hexagonal CuS [79-2321], hexagonal
Cu1.8S [47-1748], and orthorhombic Cu1.97S [20-0365]. The pattern (b) is a
mixture of Cu1.8S and CuS.
Characterization. The crystal structure of the products was
examined with a Scintag X-1 Advanced X-ray powder diffractometer (XRD, 2.4°/min, Cu KR radiation) and the morphology of
these nanocrystals was characterized using a transmission electron
microscope, JEOL 1200CX (TEM, accelerating voltage: 80 kV).
The optical properties of the products were determined with a
Varian Cary Bio50 UV-vis spectrometer. The concentrations Cwt
of the Cu2-xS NC dispersions for optical measurement were 0.02
g/L, and the solvent was toluene.
Results
Crystallinity and Morphology. Sonoelectrochemical Method.
X-ray diffraction confirmed that the sonoelectrochemical products consist of crystalline copper sulfide nanocrystals (NCs)
(Figure 1). The composition control of the Cu2-xS has been
realized by applying different potentials during the sonication.
At 0 V (vs Ag/AgCl satd KCl) the resulting product is pure
CuS (covellite). The obtained powder X-ray diffraction pattern
(Figure 1A) matches the JCPDS reference file for CuS [792321]. At 0 V potential, Cu(II) in cupric sulfate does not get
reduced. Only the active S(0) in sodium thiosulfate is reduced
to S2-. By lowering the work potential, the Cu(II) gets reduced
to Cu(I). As a result, the diginite Cu1.8S NCs were produced by
using a potential of -0.6 V, the corresponding XRD is shown
in Figure 1B. When the reduction potential was increased to
-1.2 V, a copper sulfide phase with very similar XRD to ones
that others have recently reported as Cu2S nanocrystals is
produced, whose standard data show peaks at 45.7° and
48.5°.44,45 However, upon careful research and critical review
of the previous reports, the XRD of the resulting copper sulfide
NCs, which has peaks at 46.1° and 48.6°, have to be reassigned
to Cu1.97S (djurleite), its standard data has peaks at 46.1°/46.3°
and 48.6°. The key difference in distinguishing the Cu1.97S phase
from Cu1.8S is the peak at ∼37°. On the other hand, the key
difference in distinguishing the Cu1.97S phase from Cu2S is the
peak at ∼46°. The corresponding standard JPCDS file (20-0365)
is the only one that fits the experimental data. Any Cu2S standard
JPCDS files did not match. With intermediate reducing potentials, mixtures of adjacent copper sulfides have been observed
and confirmed by the XRD pattern. This result supported the
conclusion drawn by Lambrecht et al.6 that it is actually very
difficult to make Cu2S since Cu1.97S is the thermodynamically
(45) Liu, Z. P.; Xu, D.; Liang, J. B.; Shen, J. M.; Zhang, S. Y.; Qian,
Y. T. J. Phys. Chem. B 2005, 109, 10699–10704.
Figure 3. Powder X-ray diffraction patterns of (A) hexagonal Cu1.8S
nanocrystals by thermolysis of Cu-MPA dry powder and (B) orthorhombic
Cu1.97S nanocrystals by thermolysis of dry Cu-MPA gel.
stable stoichiometry under ambient reaction condition. The
produced high temperature phases6,46 also confirm that the
Cu2-xS is formed at elevated temperature, which occurs during
the micro bubble collapse during sonication.
Hydrothermal Method. X-ray diffraction indicated that the
hydrothermal reaction of Cu-MPA with Na2S2O3 produces
crystalline copper sulfide nanocrystals (Figure 2). The compositional control of the Cu2-xS has been realized by adjusting
the reaction solution to different pH values. At pH 2 the resulting
product is mainly CuS (covellite) but one can find a weak
impurity peak, which could be identified as the peak from Cu1.8S.
The obtained powder X-ray diffraction pattern (Figure 2A) is
very similar to other reported XRD patterns of CuS nanocrystals.47,48 It matches the JCPDS data files for CuS [79-2321] and
Cu1.8S [47-1748] as mixture, as shown in Figure 2. At pH 2,
the solution is acidic, the capping compound MPA is protonated,
free Cu(II) ion is released from MPA. The active S(0) in sodium
thiosulfate is reduced to S(2-). By increasing the pH value,
the MPA is less protonated and the free Cu(II) is coordinated
(46) Chakrabarti, D. J.; Laughlin, D. E. Cu-S (Copper-Sulfur), Binary
Alloy Phase Diagrams, II Ed., 1990.
(47) Roy, P.; Srivastava, S. K. Mater. Lett. 2007, 61, 1693–1697.
(48) Dixit, S. G.; Mahadeshwar, A. R.; Haram, S. K. Colloids Surf., A
1998, 133, 69–75.
J. AM. CHEM. SOC.
9
VOL. 131, NO. 12, 2009
4255
ARTICLES
Zhao et al.
Figure 4. Evolution of powder X-ray diffraction patterns of (A) Cu1.8S
NCs synthesized via thermolysis of the Cu-MPA complex dry powder at
200 °C after different sintering times (B) Cu1.97S NCs synthesized via
thermolysis of the Cu-MPA complex gel at 200 °C after different sintering
times.
by the MPA. The lower Cu(II) ion concentration is then exposed to a relatively higher S(2-) concentration, and Cu(II) is
reduced to Cu(I).44,49 Therefore, a stronger Cu1.8S peak at 46°
was found in the XRD of Cu2-xS NCs synthesized at pH 3, as
shown in Figure 2B. There is a mixture of CuS and Cu1.8S
observed when the pH value of the reaction solution is between
3 and 5. When the solution becomes less acidic and the pH
increases to 5, Cu1.97S NCs, which match with the JCPDS 200365 file, have been produced. When the reaction solution
becomes more basic, there is less or even no Cu2-xS produced.
A possible reason is that the free Cu(II) ions are more chelated
by the deprotonated MPA and the S(0) in sodium thiosulfate is
also more difficult to release in basic solution.
Solventless Thermolysis. The resulting copper sulfides from
solventless thermolysis of the Cu-MPA complex dry powders
or dry gels in nitrogen atmosphere were also nanocrystals as
confirmed by line broadening analysis using the Scherrer
equation on the X-ray diffraction results (Figure 3). Thermolysis
of dry powders produced diginite Cu1.8S NCs, matching the
JCPDS 47-1748 file. On the other hand, thermolysis of dry gel
produced Cu1.97S NCs, which matched the JCPDS 20-0365 file.
The evolution of the crystallinity with thermolysis time can be
tracked in Figure 4. It is found that thermolysis of dry powders
needs 15 min to obtain Cu1.8S NCs while the thermolysis of
dry gel can produce the Cu1.97S NCs in 7 min. Korgel et al.23
has suggested that the mechanism of solventless synthesis of
Cu1.97S from Cu-thiolate precursor by thermolysis is a homolytic cleavage of the thiol and alkyl groups and the Cu cations
inhabit the interstitial space. At the same time, the thiols (and
perhaps the carboxyl ligands) contribute to the reduction of
Cu(II) to Cu(I). The selective production of Cu1.97S and Cu1.8S
may come from that difference between the dry gel and dry
powder of the Cu-MPA precursor. From the XRD in Figure
4, we found that the dry powder of the Cu-MPA shows some
weak peaks, which indicated that the Cu-MPA dry powder has
formed some crystal structure without sintering. On the other
hand, the gel-like Cu-MPA precursor produced Cu1.97S just as
reported by Korgel’s23 Cu-thiolate precursor, which do not
show crystallinity without sintering.
Figure 5. TEM micrographs and size histograms of CuS (A), Cu1.8S (B), and Cu1.97S (C) nanocrystals synthesized by the sonoelectrochemical method.
4256
J. AM. CHEM. SOC.
9
VOL. 131, NO. 12, 2009
Plasmonic Cu2-xS Nanocrystals
ARTICLES
Figure 6. TEM images and size histograms of of CuS NCs (A) and Cu1.97S
NCs (B) nanocrystals synthesized by applying the hydrothermal method at
pH 2 and 5, respectively.
There are a few differences between Korgel’s Cu2-xS
samples and ours that are worth considering. First, the
nanocrystals presented here are bigger than the nanocrystals
made by Korgel et al. Also, our Cu1.97S nanocrystals do not
have a disk shape. This shape was one of the factors Korgel
et al.23 considered when determining the crystal structure and
it led to the conclusion that their nanocrystals could be
hexagonal Cu2S.
Consistent with our results, Korgel could produce CuS
disks in solution, and also made some of the nonstoichiometric phases, but not Cu2S. So, there might be something
particular about the solventless reaction conditions. At least
they have a significant influence over the copper sulfide
stoichiometry.
The size and shape of the sonoelectrochemically synthesized
copper sulfide NCs are shown in Figure 5. The particle sizes of
these Cu2-xS are within a range of 5-20 nm. The particle shape
of the various copper sulfide NCs are similar. For the CuS NCs
(A), the particle sizes are 14.9 ( 7.1 nm; for the Cu1.8S NCs
(B) the particle sizes are 18.1 ( 7.9 nm; for the Cu1.97S
NCs (C) the particle sizes are 12.1 ( 4.6 nm. The average size
was counted by ImageJ software for more than 100 particles.
The size of these nanoparticles, as calculated by the Scherrer
equation on the basis of the three most intense peaks, is about
15, 16, and 18 nm for CuS (A), Cu1.8S (B), and Cu1.97S (C),
respectively, which is in reasonable agreement with the TEM
results.
The size and shape of the hydrothermally synthesized Cu2-xS
NCs are shown in Figure 6. For the CuS NCs (A) the particle
sizes are 22.3 ( 8.2 nm; for the Cu1.97S NCs (B) the particle
sizes are 30.3 ( 12.6 nm. The average size was measured by
randomly counting 200 particles in the images. The size of these
nanoparticles, as calculated by the Scherrer equation on the basis
of the three most intense peaks, is about 20 and 22 nm for CuS
(A) and Cu1.97S (B), respectively, in reasonable agreement with
the TEM results. These particles appear on the TEM grid
aggregated to each other, which may come from the use of MPA
as ligand.
The size and shape of the thermolysis synthesized copper
sulfide NCs are shown in Figure 7. For the Cu1.8S NCs (A), via
thermolysis of the dry Cu-MPA powder, the particle sizes are
(49) Silvester, E. J.; Grieser, F.; Sexton, B. A.; Healy, T. W. Langmuir
1991, 7, 2917–2922.
Figure 7. TEM images of Cu1.8S NCs synthesized by solventless
thermolysis of a dry powder (A) and Cu1.97S NCs synthesized by solventless
thermolysis of a dry gel (B).
51.2 ( 20.6 nm, as counted by ImageJ software. For the Cu1.97S
NCs (A), via thermolysis of the Cu-MPA gel, these particles
are aggregated on the TEM grid to form larger agglomerates
when the solution dries. The TEM shows that the NC size of
Cu1.8S, from thermolysis of the dry Cu-MPA powder, is smaller
than the obtained Cu1.97S NCs, from thermolysis of the
Cu-MPA gel, and the Cu1.8S NCs seem to be better dispersed.
The difference comes possibly from the crystal-like property
of the dry powder precursor, which breaks into small particles
when thermolyzed, while the Cu1.97S NCs from the gel-like
precursor tends to stay agglomerated after the transformation
from the amorphous precursor into crystalline Cu1.97S NCs. The
size of these nanoparticles, as calculated by the Scherrer equation
on the basis of the three most intense peaks, is about 26 and 30
nm for CuS (A) and Cu1.97S (B), respectively, This is significantly different from the TEM result for the Cu1.97S NCs. The
difference may come from the heavy agglomeration during the
thermolysis process.
UV-Vis Absorption Spectra. In this work we synthesized
Cu2-xS NCs with three new methods in order to compare
structural, compositional and optical properties between these
Cu2-xS NC samples. The comparison will then be extended
to the existing literature about Cu2-xS NCs. The UV-vis
absorption spectra of the differently made Cu2-xS NCs are
shown in Figure 8. By estimating a molar NC concentration
C and measuring the experimental absorbances, one can
obtain a good estimate for the extinction coefficient ε(λ)of
these NCs.
The molar concentration C of the nanocrystal dispersion is
calculated by eq 1
J. AM. CHEM. SOC.
9
VOL. 131, NO. 12, 2009
4257
ARTICLES
Zhao et al.
C ) Cwt /MNC ) Cwt /
( π6 d FN )
3
A
(1)
Cwt is the weight concentration of the nanocrystals, MNC is the
molar weight of the nanocrystals, NA is Avogadro’s constant, d
is the average diameter of the nanocrystals assuming the
nanocrystals are spherical, and F is the density of the nanocrystals assuming it is the same as bulk. The extinction coefficient
could be calculated from the measured absorbance A in eq 2
π
ε ) A d3FNA /(LCwt)
6
(
)
(2)
The density F for Cu2-xS is about 5.6 g/cm3. On the basis of
the TEM and XRD results, we assume that the diameters for
Figure 8. UV-vis spectra of Cu2-xS NCs prepared by (A) sonoelectrochemical, (B) hydrothermal, and (C) thermolysis methods.
4258
J. AM. CHEM. SOC.
9
VOL. 131, NO. 12, 2009
CuS, Cu1.8S, and Cu1.97S, synthesized by the sonoelectrochemical
method, are 15 nm. For CuS and Cu1.97S, synthesized by the
hydrothermal method, the diameters are assumed to be 25 nm,
for Cu1.8S and Cu1.97S, synthesized by the thermolysis method,
the diameters are assumed to be 30 nm. The so obtained
extinction coefficients ε(λ) from measured absorbances A for
each sample are shown in figure 8.
Sonoelectrochemical. The spectra of the sonoelectrochemically synthesized Cu2-xS NCs are shown in Figure 8A, the
absorbance of Cu1.8S and Cu1.97S NCs of ∼15 nm diameter has
a shoulder at ∼400 nm, which is similar to previously reported
10-20 nm CuS NCs,50,51 and reaches a minimum at ∼800 nm.
On the other hand, the absorbance never reaches zero intensity
but rises for longer wavelengths again, which is thought to come
from the free-carrier intraband absorbance.4 The absorbance of
CuS NCs has a broad shoulder around 450 nm and reaches a
minimum around 650 nm, but again not zero intensity. It rises
for longer wavelengths stronger than the Cu1.8S and Cu1.97S NCs
due to free-carrier absorbance.4
Hydrothermal. The spectra of hydrothermally synthesized
Cu2-xS NCs are shown in Figure 8B. The ∼25 nm Cu2-xS NCs,
hydrothermally synthesized at pH 2, were assigned to mainly
CuS and a little Cu1.8S on the basis of their XRD patterns. They
show an absorbance around 400 nm and a strong absorbance at
long wavelength, which arises on the basis of free carrier
intraband absorption.4,50,51 However, the spectrum of the
hydrothermally synthesized Cu2-xS NCs at pH 5, which is
assigned to Cu1.97S, shows a broad absorbance from 300 to 600
nm, similar to previously reported 10 nm Cu2-xS.49 Both of
these two hydrothermally synthesized Cu2-xS reach a minimum
around 600 nm then rise for longer wavelengths due to free
carrier intraband absorption.4 Generally, we find that CuS NCs
show a stronger free carrier absorbance than the other Cu2-xS
NC compositions, consistent with recently reported spectra of
CuS NCs.45,47,48
Thermolysis. The spectra of Cu2-xS NCs synthesized via
thermolysis, which were assigned to 30-70 nm Cu1.8S and
Cu1.97S, are shown in Figure 8C. The absorbance of Cu1.97S NCs
shows a broad maximum from 400 to 650 nm, which is about
50 nm red-shifted compared to Cu1.97S NCs made via sonoelectrochemical and hydrothermal methods. The absorbance of
Cu1.8S NCs also shows a broad peak from 400 to 650 nm, which
is about 70 nm red-shifted compared to Cu1.8S NCs made via
sonoelectrochemistry. Both spectra of these two thermolysisbased Cu2-xS NCs reach a minimum at around 800 nm and
then increase in intensity for longer wavelengths due to freecarrier absorbance.4 The quantum size effect for the thermolysisbased Cu1.8S and Cu1.97S NCs was less than for the corresponding NCs synthesized via the sonoelectrochemical and the
hydrothermal method, which can be rationalized on the basis
of the different size ranges.
From the UV-vis spectra for these Cu2-xS NCs, we find long
wavelength absorbance for all presented stoichiometries (x *
0). Since this NIR absorbance derives from free carriers, it
should show a stoichiometric dependence. CuS NCs show very
strong absorbance intensity, while the Cu1.8S NCs show
somewhat less absorbance and the Cu1.97S NCs show further
decreased absorbance intensity in the NIR region. We could
identify comparable trends in previously published work for bulk
(50) Ding, T. Y.; Wang, M. S.; Guo, S. P.; Guo, G. C.; Huang, J. S. Mater.
Lett. 2008, 62, 4529–4531.
(51) Zhang, H. T.; Wu, G.; Chen, X. H. Mater. Chem. Phys. 2006, 98,
298–303.
Plasmonic Cu2-xS Nanocrystals
ARTICLES
37.1°, 46.1°/46.3°, and 54.1°. While these peaks can be weak
and broad and overlooked in an experimental diffraction pattern,
it is the distance between the strong peaks at 46.1°/46.3° and
48.6°, which uniquely identifies Cu1.97S; Cu2S has strong peaks
at 45.7° and 48.5°. Comparison of these files with the literature
suggests that Cu2S is actually not produced under ambient
conditions. At least for any Cu2-xS synthesized under ambient
conditions without Ar protection, the XRD analysis of “Cu2S”
should be revisited.
Figure 9. NIR spectra of prepared Cu2-xS NCs: CuS (x ) 1), Cu1.8S (x )
0.2), and Cu1.97S (x ) 0.03) synthesized by the sonoelectrochemical method.
Relative spectral intensities are shown as measured and overlapped with
Figure 8A to obtain extinction coefficients for the NIR region.
Cu2-xS materials.48 The similarity of the spectra (for similarly
sized NCs) is quite consistent. It is shown here that the optical
properties are largely independent of the synthesis procedure.
Furthermore, a comparison between panels A and C in Figure
8 shows a size dependence effect, and a systematic size
dependence study is planned for the future. Figure 9 shows that
Cu2-xS NCs have comparable NIR absorption spectra. They
show that the NIR absorption intensity increases with increasing
x, identical to the red end of the UV-vis spectra in figure 8 in
which CuS shows the strongest NIR absorption followed by
C1.8S and then by Cu1.97S with the lowest absorption intensity
above 800 nm. Note that for perfect Cu2S the valence band is
filled and free carrier absorption would not occur. This is an
unambiguous way to distinguish Cu2S from Cu2-xS.
Discussion
Structural Properties. XRD is a powerful characterization
method to determine the crystal phase of materials. Most
reported nanocrystalline materials were determined via their
crystal structure because the XRD pattern for any of these
materials is unique and significantly different from any other
materials. However, there are a total of 86 reported XRD
patterns for Cu2-xS and the different stoichiometric copper
sulfides have very similar XRD patterns, which makes the
assignment of copper sulfide XRDs very tedious. Phase diagrams46 help additionally in assigning the investigated materials.
For example, the Cu2S (chalcocite) and Cu1.97S (djurleite) have
similar XRD pattern, and the phase space of Cu2S is very narrow
compared to Cu1.8S in the phase diagram.46 This explains some
inconsistency in recent literature between some reported assignments of Cu2-xS materials and the reported standard XRD
files. The fact that all three methods we used can produce Cu1.97S
but can not produce Cu2S under ambient conditions is consistent
with the reported result that Cu2S has a thermodynamic
instability toward copper-deficient copper sulfide phases and
favors the formation of copper vacancies even in contact with
pure copper.6 Very recently Alivisatos et al.53 prepared Cu2S
NCs under Ar protection.
It is important to point out here that, while these three XRD
patterns are very similar, they are, after careful study, distinguishable. For example the XRD of the Cu1.97S distinguishes
itself from the Cu2S XRD with very narrow double peaks at
(52) Zhuang, Z.; Peng, Q.; Zhang, B.; Li, Y. J. Am. Chem. Soc. 2008,
130, 10482–10483.
Lambrecht’s theoretical study has shown that the Cu2S will
be a monoclinic structure at low temperature,6 the hexagonal
form is stable in the range of 103-436 °C and the cubic
structure forms at high temperature.54 However, hexagonal Cu2S
was not observed in any of our attempts, based on the XRD.
Instead, the orthorhombic djurleite Cu1.97S XRD pattern was
observed. This indicated that the Cu2-xS changes its crystal
structure from Cu2S to Cu1.97S as a result of copper loss from
Cu2S and minor rearrangements of the sulfur atoms.55 An XRD
pattern consistent with standard JPCD files of Cu2S was
published recently by Alivisatos et al.53
While it is very interesting that one finds a better match with
Cu1.97S (djurleite) XRD patterns for nanomaterials that were
previously assigned as Cu2S (chalcocite),19,41,42,44,52 one should
not be too focused on these crystallographic assignments, which
probably have historical origins in mineralogy applications. It
rather shows that Cu2-xS varies with a continuous x and the
structural changes and names between different x are difficult
to determine and probably not very meaningful given that there
is a good deal of disorder in all these compositions.6 On the
other hand, knowing the 86 existing diffraction patterns for
Cu2-xS and being able to assign them to the synthesized
materials helps to determine a phase and with that a compositional range.
Optical Properties. UV-vis absorption spectra of various
Cu2-xS materials have been published earlier.11,19,45,47,51 However, they appear in the majority of recent publications without
context in size dependence or composition. Reason for this lack
of analysis of the spectra is probably the initial wrong assignment in the pioneering work as indirect band gap material, which
derives from a questionable approach to analyze the band gap
transition.4 In this early work the long-wavelength absorption
(Drude term) was truncated and neglected, which lead to
substantial arbitrariness to assign the band gap transitions as
indirect ones. Lambrecht’s work6 implied that the absorption
of hexagonal Cu2-xS is (a) direct band gap, (b) a function of x,
due to the underlying Moss-Burstein effect,56,57 in which
increasing doping (increasing x) will lead to an increase of hole
formation in the valence band and effectively lower the energy
of the lowest occupied energy level, (c) size-dependent in the
10 nm particle range. From Lambrecht’s theoretical study, the
approximate exciton mass µx for copper sulfide is 0.2 from using
eq 3 with the effective electron mass (me) and corresponding
effective hole mass (mh).6
(53) Will, G.; Hinze, E.; Rahman, A.; AbdelRahman, M. Eur. J. Mineral.
2002, 14, 591–598.
(54) Whiterside, S. L.; Goble, J. R. Can. Mineral 1986, 24, 247–258.
(55) Wu, Y.; Wadia, C.; Ma, W.; Sadtler, B.; Alivisatos, A. P. Nano Lett.
2008, 8, 2551–2555.
(56) Moss, T. S. Proc. Phys. Soc. B 1954, 67, 775–782.
(57) Burstein, E. Phys. ReV. 1954, 93, 632–633.
J. AM. CHEM. SOC.
9
VOL. 131, NO. 12, 2009
4259
ARTICLES
Zhao et al.
1
1
1
)
+
µx
me
mh
(3)
In comparison, the well-studied material CdSe has an excition
mass µx ) 0.1, and the exciton Bohr radii for CdSe is 5.3 nm.58
One can obtain an exciton Bohr radius for Cu2-xS of about 3-5
nm by using the equation for the excitonic Bohr radius aB )
ε/µx if one assumes that the relative dielectric constant ε for
copper sulfides is about 10-15. This will likely be a function
of the composition in Cu2-xS and a dielectric constant 10-15
seems reasonable for small x compared to other semiconductor
materials.58 The size effect range for CdSe QD is from 1-6
nm. Similarly, we expect to see a strong quantum size effect
for Cu2-xS in the sub-10 nm range. In agreement with this
analysis, several reported Cu2-xS NCs show blue-shifted band
gap absorption compared to bulk material, while their sizes are
in the 10 nm region.45,59,60 However, the dielectric constant may
even be larger than these regular values because of the free
carriers in these nonstoichiometric Cu2-xS, which leads to larger
Bohr radii and broader quantum size ranges. For example PbTe
has a dielectric constant of ∼100 due to its narrow band gap
energy and a Bohr radius of ∼50 nm.61,62 The dielectric constant
of Cu2-xS could well depend on x and reach high values for
compositions with larger x values. This would accordingly
increase the exciton bohr radii by an order of magnitude. Indeed,
the large blue-shifts observed in Figure 8 and the consistently
underestimated band gap energies by theory6 strongly hint
toward large exciton Bohr radii and dielectric constants for
Cu2-xS nanomaterials. To the best of our knowledge, these basic
properties are to date undetermined for copper deficient Cu2-xS.
All UV-vis spectra of the synthesized copper sulfide NCs
show a broad blue-shifted peak around 400-600 nm, indicating
size quantization effects, which has been reported earlier for
the case of Cu2-xS NCs.5,63 On the other hand, the UV-vis
spectra of Cu2-xS NCs do not show strong compositional
dependence as was reported for bulk copper sulfide.37 A possible
reason for this is that the quantum size effect in these Cu2-xS
prevails over the band gap changes due to the Moss-Burstein
effect.56,57 This can be confirmed by the fact that a more
significant blue-shift is observed in the 5-20 nm Cu2-xS NCs
synthesized via sonoelectrochemical and hydrothermal methods
compared to the 20-70 nm Cu2-xS NCs, which were synthesized via thermolysis. The obtained experimental extinction
coefficient ε for these synthesized Cu2-xS NCs in the region of
107-108 M-1 cm-1confirms that these materials should all be
of direct band gap character.
Plasmonic NIR Absorption. The NIR absorption of these
Cu2-xS NCs can be interpreted in terms of valence band free
carriers, which are essentially metallic in character and give
rise to a Drude term with corresponding NIR plasmon absorption
(Figure 9). Metals provide the best evidence of plasmons,
because they have a high density of free electrons. In the
presented Cu2-xS nanoparticles, the majority charge carriers are
positive holes. The frequency of a plasma-wave oscillation is
(58) Fu, H. X.; Wang, L. W.; Zunger, A. Phys. ReV. B 1999, 59, 5568–
5574.
(59) Roy, P.; Srivastava, S. K. J. Nanosci. Nanotechnol. 2008, 8, 1523–
1527.
(60) Behboudnia, M.; Khanbabaee, B. J. Cryst. Growth 2007, 304, 158–
162.
(61) Kanai, Y.; Shohno, K. Jpn. J. Appl. Phys. 1963, 2, 6–10.
(62) Wise, F. W. Acc. Chem. Res. 2000, 33, 773–780.
(63) Haram, S. K.; Mahadeshwar, A. R.; Dixit, S. G. J. Phys. Chem. 1996,
100, 5868–5873.
4260
J. AM. CHEM. SOC.
9
VOL. 131, NO. 12, 2009
determined by the carrier density and the energy of the
corresponding plasmon is its frequency multiplied by Planck’s
constant (h).
From Mie theory, the extinction coefficient for N particles
of volume V is given by the following equation.64,65
κext(λ) ) 18πNVε3/2
m
ε2(ω)
λ[(2εm + ε1(ω))2 + ε2(ω)2]
(4)
Where ε1(ω) and ε2(ω) are the real and imaginary parts of the
dielectric constant, respectively, εm is the dielectric constant of
the surrounding medium.
From the Drude theory, ε1(ω) and ε2(ω) is given by following
equations
ε1(ωp) ) εcore -
ne2τ
4πne2τ2
4π
ε (ωp) )
2 2 2
m*ωp 1 + (ω τ)2
m*(1 + ωpτ )
p
(5)
Where ωp is the plasma frequency, e is the elementary charge,
n is the free carrier (hole for the Cu2-xS) concentration, ε is the
dielectric constant, m* is the charge carrier effective mass and
τ is the average relaxation time.
The plasmon resonance condition is 2εm + ε1 ) 0 if ε2(ω) is
small or only weakly dependent on ω.65 One can obtain ωp from
eq 5
ω2p )
1
4πne2
4πne2
- 2 ≈
m*(2εm + εcore)
m*(2εm + εcore)
τ
(6)
It is shown in eq 6 that the plasmon frequency ωp increases
with free carrier concentration (n1/2). In other words, the
resonance wavelength λ will decrease with increasing x in
Cu2-xS, as is observed in Figure 9.
At plasmon resonance, 2εm + ε1 ) 0,65 the extinction
coefficient eq 4 can be simplified by using eqs 5 and 6 and
ωp2τ2 . 1 for optical to NIR frequencies
κext(λ) )
3
18NVε3/2
18πNVε3/2
(4πne2)1/2
m m*ωpτ
m τ
)
2
*1/2
λ
λ
4πne
m (2εm + εcore)3/2
(7)
From eq 7, one finds that the extinction coefficient κext(λ) at
wavelength λ increases with the free carrier concentration n1/2
(i.e., increasing x) assuming constant N and V for the investigated Cu2-xS NPs samples. This is also directly observed in
Figure 9. For perfect nonoxidized Cu2S NPs, such a plasmon
absorbance would not exist, which is supported by very recent
work of Alivisatos et al.55 and we could confirm that the
absorbance of so prepared Cu2S NCs starts to show long
wavelength absorbance after the Cu2S NCs were exposed to
air.
Conclusion
In summary, we have synthesized a series of compositionally
controlled Cu2-xS nanocrystals via three novel different chemical
procedures using sonoelectrochemical, hydrothermal, and thermolysis methods. The range of compositions of the Cu2-xS
nanocrystals ranged from CuS (covellite) to Cu1.97S (djurleite)
through adjusting the reduction potentials from 0 to -1.2 V in
(64) Papavassiliou, G. C. Prog. Solid State Chem. 1980, 12, 185–286.
(65) Link, S.; El-Sayed, M. E. J. Phys. Chem. B 1999, 103, 8410–8426.
Plasmonic Cu2-xS Nanocrystals
the sonoelectrochemical method, from CuS to Cu1.97S (djurleite)
by adjusting the pH values from 2 to 5 in the hydrothermal
method and from Cu1.8S to Cu1.97S in the thermolysis method.
On the basis of these results, the Cu2S is found to be
thermodynamically unstable under ambient condition, compared
to the copper deficient Cu2-xS phase. The UV-vis interband
spectra of the synthesized Cu2-xS NCs show a blue shift in the
absorption compared to bulk (a) due to a quantum size effect
and (b) a Moss-Burstein effect caused by the copper deficiency
in Cu2-xS. The intensity of the observed spectra show that all
studied Cu2-xS nanocrystals should be of direct band gap type.
ARTICLES
The observed properties of the synthesized Cu2-xS nanocrystals
are in accordance with recent calculations by Lambrecht et al.
The NIR spectra of these Cu2-xS NCs show a plasmonic
absorption, which is carrier concentration dependent.
Acknowledgment. C.B. gratefully acknowledges financial support from NSF (CHE-0239688), NIRT (0608896), and ACS PRF
(45359-AC10). J.J.Z. is financially supported by the National
Natural Science Foundation of China (20635020, 20325516).
JA805655B
J. AM. CHEM. SOC.
9
VOL. 131, NO. 12, 2009
4261